Perceptual decision-making involves multiple cognitive components and diverse brain regions. To perform a perceptual decision, an individual must process an incoming sensory percept, retain this information in short- term memory, and choose an appropriate motor action. Research using delayed-response tasks in nonhuman primates has revealed that sensory and choice information is distributed across a hierarchy of cortical areas, with task-relevant information flowing from sensory to association to motor regions. However, a mechanistic understanding of how circuits in these regions transform and maintain information during such tasks is lacking, due to limited ability to identify and manipulat specific circuits in the primate brain. By developing a memory- guided task for head-fixed mice, we intend to leverage the genetic tractability of the mouse to address these questions. We have developed a perceptual decision task for mice that involves separate sensory, memory, and action epochs. Using large-scale population calcium imaging (Aim 1), we can simultaneously measure the activity of 1000+ neurons during the task, and across multiple brain regions (visual, parietal, and frontal motor cortex). This will allow us to record how neural activity in different cortical areas correlates with different epochs of the task. Our preliminary results indicate a diversity of different response types in each of the three areas studied, including delay-period activity in a large proportion of parietal and motor cortical neurons. These huge and complex data sets require us to employ new statistical methods (Aim 2) to analyze cell-type-specific and region-specific population activity patterns. In collaboration with Emery Brown, we will use state-space approaches to infer how single cells and cortical areas encode information about the task. To investigate the specific circuits and projection pathways underlying the task (Aim 3), we will use retrograde tracers such as rabies virus (in collaboration with Ian Wickersham) to label neurons that project to a particular brain region, or even to a single task-responsive neuron, and measure their functional role during the task. In collaboration with Kwanghun Chung, we will then use CLARITY for multiple-protein immunostaining of the entire brain. These techniques in combination will allow us to link the molecular identity and connectivity profile of each neuron with its functional role in the task. Finally, we plan to test the causal role of these brain regios and circuits using novel ontogenetic tools (Aim 4). Using transgenic mice that express ChR2 in inhibitory neurons, we will transiently inactivate each brain region during specific epochs of the task. This will allow us to determine the necessity and time course of involvement of each brain region. We will lastly manipulate the activity of anatomically-defined and computationally-identified subsets of neurons within each brain region, to determine whether specific subpopulations play a causal role in behavior. By integrating a wide range of cutting-edge experimental and computational tools, and assembling a collaborative team with multidisciplinary expertise, we hope to transform understanding of the neural substrates underlying memory-guided perceptual decisions.